Load EDT form elements on demand

packages/one-c-normalizer/src/one_c_normalizer/__init__.py

@@ -1,7 +1,7 @@
 from __future__ import annotations
 
 import hashlib
-from pathlib import Path
+from pathlib import Path, PurePosixPath
 import re
 import xml.etree.ElementTree as ET
 
@@ -362,13 +362,13 @@ def _walk_xml_objects(
         if right is not None:
             result.append(right)
     elif object_kind is not None:
-        name = _xml_name(element)
+        name = _xml_name(element, source_path=source_path)
         if name:
             xml_object = OneCXmlObject(
                 source_path=source_path,
                 object_kind=object_kind,
                 name=name,
-                qualified_name=_xml_qualified_name(element, name, object_kind, parent_qualified_name),
+                qualified_name=_xml_qualified_name(element, name, object_kind, parent_qualified_name, source_path),
                 attributes=_xml_attributes(element),
             )
             result.append(xml_object)
@@ -1332,7 +1332,7 @@ def _xml_type_name(element: ET.Element) -> str:
     return ""
 
 
-def _xml_name(element: ET.Element) -> str:
+def _xml_name(element: ET.Element, *, source_path: str = "") -> str:
     for key in ("name", "Name", "Имя"):
         if key in element.attrib:
             return element.attrib[key]
@@ -1340,6 +1340,9 @@ def _xml_name(element: ET.Element) -> str:
         if _local_name(child.tag).lower() in {"name", "имя"} and child.text:
             return child.text.strip()
     tag = _local_name(element.tag).lower()
+    edt_form = _edt_form_context_from_path(source_path)
+    if tag == "form" and edt_form is not None:
+        return edt_form[0]
     fallback_keys = {
         "urltemplate": ("template", "url", "path", "Шаблон", "URL"),
         "urltemplates": ("template", "url", "path", "Шаблон", "URL"),
@@ -1388,6 +1391,7 @@ def _xml_qualified_name(
     name: str,
     object_kind: str,
     parent_qualified_name: str | None,
+    source_path: str = "",
 ) -> str:
     for key in ("qualifiedName", "QualifiedName", "ПолноеИмя"):
         if key in element.attrib:
@@ -1395,6 +1399,10 @@ def _xml_qualified_name(
     for child in _xml_property_children(element):
         if _local_name(child.tag).lower() in {"qualifiedname", "полноеимя"} and child.text:
             return child.text.strip()
+    if object_kind == "FORM" and parent_qualified_name is None:
+        edt_form = _edt_form_context_from_path(source_path)
+        if edt_form is not None:
+            return edt_form[1]
     if parent_qualified_name:
         if object_kind in _ROOT_METADATA_OBJECT_KINDS and object_kind not in {"PROJECT", "ROLE"}:
             prefix = _QUALIFIED_PREFIX_BY_KIND.get(object_kind, object_kind)
@@ -1407,6 +1415,48 @@ def _xml_qualified_name(
     return name
 
 
+_EDT_OWNER_PREFIX_BY_DIRECTORY = {
+    "AccountingRegisters": "РегистрБухгалтерии",
+    "AccumulationRegisters": "РегистрНакопления",
+    "BusinessProcesses": "БизнесПроцесс",
+    "CalculationRegisters": "РегистрРасчета",
+    "Catalogs": "Справочник",
+    "ChartsOfAccounts": "ПланСчетов",
+    "ChartsOfCalculationTypes": "ПланВидовРасчета",
+    "ChartsOfCharacteristicTypes": "ПланВидовХарактеристик",
+    "DataProcessors": "Обработка",
+    "DocumentJournals": "ЖурналДокументов",
+    "Documents": "Документ",
+    "Enums": "Перечисление",
+    "ExchangePlans": "ПланОбмена",
+    "ExternalDataSources": "ВнешнийИсточникДанных",
+    "InformationRegisters": "РегистрСведений",
+    "Reports": "Отчет",
+    "Tasks": "Задача",
+}
+
+
+def _edt_form_context_from_path(source_path: str) -> tuple[str, str] | None:
+    if not source_path or PurePosixPath(source_path).name.casefold() != "form.form":
+        return None
+    parts = PurePosixPath(source_path).parts
+    try:
+        forms_index = parts.index("Forms")
+    except ValueError:
+        forms_index = -1
+    if forms_index > 1 and forms_index + 1 < len(parts):
+        owner_directory = parts[forms_index - 2]
+        owner_name = parts[forms_index - 1]
+        form_name = parts[forms_index + 1]
+        owner_prefix = _EDT_OWNER_PREFIX_BY_DIRECTORY.get(owner_directory)
+        if owner_prefix:
+            return form_name, f"{owner_prefix}.{owner_name}.{form_name}"
+    if len(parts) >= 3 and parts[-3] == "CommonForms":
+        form_name = parts[-2]
+        return form_name, f"ОбщаяФорма.{form_name}"
+    return None
+
+
 def _xml_attributes(element: ET.Element) -> dict:
     attributes = dict(element.attrib)
     for key, value in element.attrib.items():
@@ -1445,10 +1495,18 @@ def _xml_nested_text_value(element: ET.Element) -> str:
         return localized.get("ru") or localized.get("ru_RU") or next(iter(localized.values()))
     if _local_name(element.tag).lower() == "value":
         return _element_text_content(element)
+    path_segments = [
+        _element_text_content(child)
+        for child in element
+        if _local_name(child.tag).lower() in {"segment", "segments", "pathsegment"}
+    ]
+    path_segments = [value for value in path_segments if value]
+    if path_segments:
+        return ".".join(path_segments)
     values = [
         _element_text_content(child)
         for child in element
-        if _local_name(child.tag).lower() in {"value", "text", "строка", "представление"}
+        if _local_name(child.tag).lower() in {"value", "text", "строка", "представление", "caption"}
     ]
     values = [value for value in values if value]
     if values:

packages/semantic-kernel/tests/test_xml_indexing.py

@@ -373,6 +373,38 @@ def test_index_project_extracts_managed_form_items_without_layouts(tmp_path: Pat
     assert not any(element.name == "ПечатнаяФорма" for element in form.elements)
 
 
+def test_index_project_extracts_edt_form_items_from_form_file_path(tmp_path: Path):
+    form_dir = tmp_path / "src" / "Catalogs" / "ВидыЗаказовПокупателей" / "Forms" / "ФормаСписка"
+    form_dir.mkdir(parents=True)
+    (form_dir / "Form.form").write_text(
+        """
+<form:Form xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns:form="http://g5.1c.ru/v8/dt/form">
+  <items xsi:type="form:Table">
+    <name>Список</name>
+    <dataPath xsi:type="form:DataPath">
+      <segments>Список</segments>
+    </dataPath>
+    <items xsi:type="form:FormField">
+      <name>Наименование</name>
+      <dataPath xsi:type="form:DataPath">
+        <segments>Список.Description</segments>
+      </dataPath>
+    </items>
+  </items>
+</form:Form>
+""",
+        encoding="utf-8",
+    )
+
+    snapshot = index_project(tmp_path, project_id="edt-form-items")
+
+    form = next(item for item in form_semantics(snapshot) if item.form.name == "ФормаСписка")
+    assert form.form.qualified_name == "Справочник.ВидыЗаказовПокупателей.ФормаСписка"
+    assert [element.name for element in form.elements] == ["Список", "Наименование"]
+    assert form.elements[0].attributes["control_kind"] == "Table"
+    assert form.elements[1].attributes["dataPath"] == "Список.Description"
+
+
 def test_index_project_links_form_events_to_handlers(tmp_path: Path):
     xml = tmp_path / "form.xml"
     xml.write_text(